Typically, a rotary ball bearing for a shaft or the like is constructed with two rings, namely an inner ring mounted to the shaft and an outer ring mounted to a stationary support or housing. The two rings define opposing races and are separated by a circular array of balls. Relative rotation of the rings results in the rolling of the balls along the races in the rings. In order to reduce rolling friction and to minimize wear of the bearing parts, the bearing is normally lubricated with a viscous lubricant such as oil or grease which occupies the spaces between the balls and the walls of the races. If the bearing is pre-loaded axially so that the internal clearances between the parts of the bearing are more or less removed, the balls are constrained against "skidding" in their races during normal operation of the bearing. In other words, the initial relationship between the balls and the races is fixed. Thus, if one bearing ring is rotated relative to the other back and forth through a small angle in reciprocation, each ball of the bearing rolls over a definite portion of the race in each ring and is constrained to roll over these same small areas of the races as long as the bearing remains in use.
FIG. 3 shows a conventional rotary bearing as might be used in a reciprocating device such as a galvanometer. The bearing comprises an inner ring 2 and a concentric outer ring 4, the two rings being separated radially by balls 6 which roll in races defined by the opposing surfaces of the rings. In cross section, those races may have cylindrical or, more often, elliptical curvature so stresses are highest where the radially inner and outer poles of the balls contact the bottoms of the races. A lubricant is invariably provided between the rings to minimize wear of the bearing parts. In a typical application, e.g., a shaft bearing for a galvanometer, the inner ring 2 may be angularly reciprocated relative to outer ring 4 about an axis A through a small angle .theta. of, say, 0.degree.-15.degree.. Since the balls 6 roll along the same small segments of the races, after only a few minutes of operation, the lubricant present in the reciprocating ball bearing is squeezed out of the high-pressure regions between the balls and the races, primarily at the ball poles which contact the bottoms of the races. After only a relatively few reciprocations of the bearing ring 2 through an angle .theta. of, say, 8.degree., small lubricant "hills" 8 build up in the races at the extreme ends of each ball's excursion along the races. Since the angle .theta. through which the bearing rings 2 and 4 rotate relatively is always the same, there is no mechanism to return the lubricant 8 to the high-pressure areas of the bearing between the ball poles and bottoms of the races where it is needed. Resultantly, those areas tend to wear excessively.
If now the bearing rings 2 and 4 are caused to rotate relatively through a larger angle .theta. of, say 10.degree. for one reason or another, the lubricant hills 8 deposited as aforesaid will increase the torque required to rotate the bearing momentarily. Of course, once the new angular excursion of the bearing ring 2 is established, the original lubricant hills 8 will disappear and reform on the races at the new extremes of ball travel. This process goes on continuously for the life of the lubricant in a bearing used in a random reciprocating application and may not be noticed. Eventually however, all of the excess lubricant is "parked" at the extremes of the allowed bearing ball travel and thereafter has no great affect on day-to-day bearing operation. However, as noted above, the lubricant is not present in the high-pressure areas of the bearing where it is most needed. Consequently, excessive wear occurs at those locations so that the bearing has a relatively short useful life. In fact, the aforesaid lubricant parking phenomenon is the main cause of bearing failure in most reciprocating bearings since, unless one bearing ring can rotate completely around relative to the other, the excess lubricant is effectively lost to the lubricant replenishment process. This is why many failed bearings seem to have plenty of lubricant remaining in them. The lubricant is in fact there, but it is not available at the high-pressure areas of the bearings where it is most needed.